some can help out in making a popper CSD waterfall ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Question ! i was tryin to measure this weird driver YouTube

i am no expert in taking measurements. so i compared some settings from cumulative spectral decay measurements from this video of new record day alias GR audio. since i want to to make those plots to and be able to help me in maybe changing stuff for the better.

here are the examples i used YouTube

And made this measurement. and a picture of my settings , if you see something completely out of wack let me know.

Window used 4ms

CSD on

time window 4ms

rise time 0.6.

in the video they used 0.58ms not sure what that matters.

I adjusted the bottom floor to -25dB from the overall spl not the peak at 3Khz just as i see in the clio measurements from GR. although they do use the peak , then there would be less detail, i thought this might be more fair.

Ii measured at 0.25 meter and compensated the SPL for it. so the spl seen is actual spl at 1 meter. to get my room and noise floor down. (i know for such a huge thing also not ideal in frequency response to measure that close, but thats not what i am looking for now)

Driver tested is the big ribbon.
 

Attachments

  • spectral.jpg
    spectral.jpg
    356.9 KB · Views: 239
  • Spectral settings.jpg
    Spectral settings.jpg
    290.3 KB · Views: 240
  • 20191110_184607.jpg
    20191110_184607.jpg
    1,019.2 KB · Views: 244
  • 20191110_183133.jpg
    20191110_183133.jpg
    961 KB · Views: 244
Last edited:
I look forward to some expert advice here too

Looks to me like theres some energy storage between 3-4 khz, 8k, 10k, quite typical, BUT what confuses me is how quickly its down to nothing. Doesnt seem possible that area down to nothing that quick ??

I personally have been using waterfalls with mic about 1/4 inch from diaphragm. Gets the room and diffraction out of it. May not be the "correct" way BUT I get consistent repeatable results and find it much easier to find the problem areas.

There are problems with this methode BUT the areas I find that need work are exposed easily this way.
 
yes very acceptable, thats what confuses me. I have never seen a construction like that where the wiggles decay that fast. But im usually looking at much less wide "ribbons" so it could be your larger planers structure suppresses these better than a ribbon only 1" wide??
I would expect to see some much longer "tails" in the csd so I wonder about the measurement. Usually, not sure this is "correct", but usually I have to do these measurements with mic only 1/4 inch from diaphragm and move around over its surface to see what going on.
 
Last edited:
yes very acceptable, thats what confuses me. I have never seen a construction like that where the wiggles decay that fast. But im usually looking at much less wide "ribbons" so it could be your larger planers structure suppresses these better than a ribbon only 1" wide??
I would expect to see some much longer "tails" in the csd so I wonder about the measurement. Usually, not sure this is "correct", but usually I have to do these measurements with mic only 1/4 inch from diaphragm and move around over its surface to see what going on.

1/4 inch. That close. I think it also has to do with settings of the csd. I used the same i believe as i seen in other measurements . (Might still made a mistake somewhere)
Measuring such a huge thing 1/4 will give a scewed result, allot of low end and prob higher spl then actually is true. And it might not run parallel ringing vs fundamental . Because of proximity. Just an idea.

Well the reason the traces die out so fast might be the because i patched up a fault in the design at first. The mylar in between the traces created really anoyong buzzing sounds. And in this design i patched up as last resort before it would end up in the garbage bin... like well lets try it since i will trow it out anyway. Cant yell yet exactly what. Since i want to be sure first. I lost thr measurement before the patch

But a normal ribbon might be much lighter. But les damped? This panel feels and sound like a soft floppy thing witch does not habe much noise of it own when handled.

Im gone try a new one with a few pieces of thicker foil spaced out unevenly horizontally . I hoppe to stabelize the sides more compared to thr muddle part , since thst part is held nucely because it is supported by eveb more ribbon. The sides only on one side. Not sure if i explain that well...
 
Well im certainly no expert here BUT I have been at 1/4 inch to get everything room and diffraction related out of the measurement. I cannot get consistent measurements for decay with mic farther away.

Also seems that to get well below 1 ms almost all the way down to 1K is a seriously good result, better than I have ever seen in similar constructions.
 
Well im certainly no expert here BUT I have been at 1/4 inch to get everything room and diffraction related out of the measurement. I cannot get consistent measurements for decay with mic farther away.

Also seems that to get well below 1 ms almost all the way down to 1K is a seriously good result, better than I have ever seen in similar constructions.

well lets hope... but thats the thing why i ask all around ! i dont want to be happy about a result and then somone undo it all since i forgot something or something :) haha

here is a spectrogram. im having a hard time to even read those since i dont know where to look for hopefully anyone here is more into them.. since aprantly everyone is looking for something else in a measurement .... :(


at least the 3.8Khz peak is visible :)


i hope the elder die hard measurement and calculation guru's might be able to shine a light on those ? im not one of them ... i just make stuff haha
 

Attachments

  • spectrogram.jpg
    spectrogram.jpg
    208.2 KB · Views: 217
  • spectrogram settings.jpg
    spectrogram settings.jpg
    352.5 KB · Views: 73
Last edited:
the spectrogram has same info as the waterfall I believe, just another way to "see it"

the 600-1500 stuff is typical and doesn't look bad. In fact it looks like an UN tensioned diaphragm to me. Put a little tension on it and those areas will amp up quite a bit. you could try some restrictive "screen" tacked on back of magnet structure to damp this area some
 
Last edited:
the spectrogram has same info as the waterfall I believe, just another way to "see it"

the 600-1500 stuff is typical and doesn't look bad. In fact it looks like an UN tensioned diaphragm to me. Put a little tension on it and those areas will amp up quite a bit. you could try some restrictive "screen" tacked on back of magnet structure to damp this area some


I was having that duscusdion eith some , that in my believe it is the same... some even mentioned i needed to have a mix reavhing up to 50khz to say something about the plot.... as far as i know its not harmonicso 20khz will do ... but who knows.

Wel yeah i could damp it some more altrough there is a some already there. If i would make the low end really importand i maybe even add front magnets. Since it will hurt the top end... witch is not useable because of the size anyways.
 
heres a small ribbon

fist with small tension, 2nd with a little more tension

yeah so the little more gives other problems it extend more in the upper range and less in the lower. hmm

on the other hand effciency went up to and so did the ringing. also i wonder if 200 hz is measureable with only gating ? most gating settings are like 4 ms

if im correct that would be 688hz ? or do i have to deivde that ones making it 244Hz. hmm since out of the gating range things look like **** anyways since you see all reflections

it is not actual gating but just the time window , but i think that does the same in a spectral
 
Last edited:
@ Wrinex,

CSD and spectrogram plots do require more care and experimentation to understand what you have. In particular, the uncertainty principle is at play in that you cannot calculate and display high resolution in time and frequency simultaneously. For example if you adjust the appodising window rise time to a lower value, you will pick up resolution in time but lose resolution in frequency. Alternatively, increased rise times will increase the frequency resolution, but resolution in time suffers. Somewhere I've got a set of plots from an AES paper illustrating this...will see if I can locate and post. Obviously manufacturers can take advantage of this to manipulate plots in brochures to put the best spin on a product. Often it is best to create a set of plots with a range of setting to see the 2D trends in frequency and time to better understand the severity of delayed resonance at different frequencies. You will also notice that traditional CSD plots have better resolution at high frequencies than low frequencies. Some software packages like ARTA offer burst decay analysis which provides uniform resolution at all frequencies.

Spectrogram plots are also usually generated so as to have uniform frequency resolution. But again, changing the settings for increased frequency resolution, and the time resolution suffers. Attached is an example taken from a Audiomatica presentation. The increased Q setting in CLIO, increases frequency resolution but decreases time resolution. More information posted here if interested: Electrostats vs conventional drivers

Can you export your impulse measurement as *.wav file and post it? I would be interested to play around with it to see what delayed resonance information can be extracted. As lowmass mentioned, a nearfield measurement will often let you see delayed resonance behavior in ribbons more easily because of the better signal to noise ratio it provides.
 

Attachments

  • Spectrogram_Q.png
    Spectrogram_Q.png
    276.7 KB · Views: 145
@ Wrinex,

CSD and spectrogram plots do require more care and experimentation to understand what you have. In particular, the uncertainty principle is at play in that you cannot calculate and display high resolution in time and frequency simultaneously. For example if you adjust the appodising window rise time to a lower value, you will pick up resolution in time but lose resolution in frequency. Alternatively, increased rise times will increase the frequency resolution, but resolution in time suffers. Somewhere I've got a set of plots from an AES paper illustrating this...will see if I can locate and post. Obviously manufacturers can take advantage of this to manipulate plots in brochures to put the best spin on a product. Often it is best to create a set of plots with a range of setting to see the 2D trends in frequency and time to better understand the severity of delayed resonance at different frequencies. You will also notice that traditional CSD plots have better resolution at high frequencies than low frequencies. Some software packages like ARTA offer burst decay analysis which provides uniform resolution at all frequencies.

Spectrogram plots are also usually generated so as to have uniform frequency resolution. But again, changing the settings for increased frequency resolution, and the time resolution suffers. Attached is an example taken from a Audiomatica presentation. The increased Q setting in CLIO, increases frequency resolution but decreases time resolution. More information posted here if interested: Electrostats vs conventional drivers

Can you export your impulse measurement as *.wav file and post it? I would be interested to play around with it to see what delayed resonance information can be extracted. As lowmass mentioned, a nearfield measurement will often let you see delayed resonance behavior in ribbons more easily because of the better signal to noise ratio it provides.

THanks once more Bolserst, i hoped you would chip in to be honest :)

Burst decay was mentioned to me by a fellow person on facebook. but i dont have arta :( ill try to extract the .wav and try to do some reading. People also mentioned you need a mic that can at least go much higher in frequency (min does 20khz) response to be able to measure this correct.. it sound weird to me is that true ? im not measuring harmonics or something. i measured at around 25 cm or something. to get the background noise down.
 

Attachments

  • mega ribbon.zip
    952.4 KB · Views: 38
they idea of the thing was making an alternative to fullrange planars. like i did before, but i noticed the sound down low in the ribbon version sound more pleasing maybe because of the lack of an insane resonance ;) but i only know for sure if i make one that has 5 times the surface area or more.
 
they idea of the thing was making an alternative to fullrange planars. like i did before, but i noticed the sound down low in the ribbon version sound more pleasing maybe because of the lack of an insane resonance ;) but i only know for sure if i make one that has 5 times the surface area or more.

So a few years ago I was working on a nearly identical driver to what ya got there in the development loop. However it was a curved diaphragm, something quite easy to do with the diaphragm corrugated like that,. The idea was to make a small plainer ( about 5 x 5 inch) that could goto 500 hz AND have better horizontal dispersion than a flat diaphragm of same size.

I worked that thing till I was crazy in attempt to make an alternative to the sound of a good free swinging ribbon. It just seemed that no matter what I did it just never sounded quite as relaxed and convincing as the ribbon. It got quite good BUT just not that last few percent that makes something truly great.

I was disappointed. The planer was good and reliable and and ..... BUT every time I went back the the free swing ribbon I knew right away it wasnt as top class.

From there I went back to flat and made many different protos to improve the sound. Most of these were simply going to smaller and smaller magnets to get more and more small foil traces( each one driven by its own magnet run so no multi runs of foil between large magnets spread out too much)
the idea was to get as uniform a drive as possible.
The sound got better BUT it never quite reached the same level fidelity as the free swing ribbon.

In the end it seems to me that the resonances that build in a planer tend to be a bit more edgy than those that build in a free swing ribbon. I suspect the ribbons very low resonance freq and the more benign nature of those resonances are tough to beat.

I did eventually make big curved ones as full range. The bass didn't work in a curved diaphragm BUT in flat it was pretty good , loved having no crossover. BUT again when I went back to a two way with the ribbon going down to 500 I thought the ribbon was more convincing.
 
Last edited:
So a few years ago I was working on a nearly identical driver to what ya got there in the development loop. However it was a curved diaphragm, something quite easy to do with the diaphragm corrugated like that,. The idea was to make a small plainer ( about 5 x 5 inch) that could goto 500 hz AND have better horizontal dispersion than a flat diaphragm of same size.

I worked that thing till I was crazy in attempt to make an alternative to the sound of a good free swinging ribbon. It just seemed that no matter what I did it just never sounded quite as relaxed and convincing as the ribbon. It got quite good BUT just not that last few percent that makes something truly great.

I was disappointed. The planer was good and reliable and and ..... BUT every time I went back the the free swing ribbon I knew right away it wasnt as top class.

From there I went back to flat and made many different protos to improve the sound. Most of these were simply going to smaller and smaller magnets to get more and more small foil traces( each one driven by its own magnet run so no multi runs of foil between large magnets spread out too much)
the idea was to get as uniform a drive as possible.
The sound got better BUT it never quite reached the same level fidelity as the free swing ribbon.

In the end it seems to me that the resonances that build in a planer tend to be a bit more edgy than those that build in a free swing ribbon. I suspect the ribbons very low resonance freq and the more benign nature of those resonances are tough to beat.

I did eventually make big curved ones as full range. The bass didn't work in a curved diaphragm BUT in flat it was pretty good , loved having no crossover. BUT again when I went back to a two way with the ribbon going down to 500 I thought the ribbon was more convincing.

Hmm i got rather ok result using multitraces into one gap, as long as foil is used. this ribbon essentially does the same. it has 10 traces running from center of the magnet (slightly off center ofc) to the center of the next magnet. int this design i used a way to beefy gap16mm or something. so field is not soo strong and or uniform. but what i did notice and i might be crazy. these blocks of ferite although weak, look to extend further fromt the mangets then there waaay smaller but stronger neodymium counter parts.. i looked if there is such a thing but could not find any proof of that. but stuff gets drawn into them at bigger distances. not if you compare the same size to the same size neo ofc. but that would be a deathtrap and so expensive it would be able to buy 10 magnets and im broke haha , wait a minute... i already am :)
 
ceramics have a better "reach out" than neos

size for size you get a more uniform field strength with ceramics

is it ?? since thats what i thought when buying them, then i made a video about it and thought...... wait lets double check that before i say something stupid. then i could not find any proof of that although i thought i read it somewhere (does not mean its true but at least more info then nothing :))

then i tested it for myself and though hmmm it does seem like they do but i might be wrong because its purely a size thing ? maybe. and i would enver handle neos of the same size.

well in the end they are cheaper and reach further then neos with equivalent pulling force. i cant get an even field with neos that have the same pulling force over the same distance.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.